Privacy, data and legal issues: challenges the BBC faced in developing its Own It app
The BBC’s Jon Howard discussed the research and development that went into creating their app to help children navigate the online world.
Companies operating in areas where their audiences might be exposed to online harms are frequently met with challenges surrounding privacy, legality of content, data and commercial considerations. This can be hard for any business, in particular SMEs, and especially in the current climate where business resource is scarce.
In the latest session hosted by the Safety Tech Innovation Network, on 15 January 2021, Jon Howard, Executive Product Manager, BBC Design and Engineering, discussed how the corporation created the Own It app, as part of its commitment to supporting young people in today’s challenging digital environment.
Watch the recording of the event.
What is the BBC Own It app?
There are two parts to the app. The first is a keyboard plug-in that, through machine learning, analyses information that a child types. After building a picture of a child’s activity, the app then recommends content that might be helpful or acts as an intervention before the young person shares personal information.
Secondly, children are encouraged to “self-report” how they are feeling through the app by selecting an emoji that best represents their mood, the app will then recommend content to support with their wellbeing.
Read more about the development of the app.
Discover, Define, Develop and Deliver
In his talk, Jon walked the audience through the four stages of app development and discussed the challenges that were overcome along the way:
- Discover: This stage was crucial in understanding how to best engage the users of the app and ensure that the team was designing for audience needs. From feedback provided by children, it was clear that a monitoring app was not wanted but, instead, the BBC’s app should provide young people with the opportunity to access a safe personal space
- Define: The app was the BBC’s first innovation in the responsible use of AI. It was at this stage, when the team focused on how the app was going to use data transparently, that they adopted a privacy-first approach to the app
- Develop: This was the BBC’s first machine learning partnership, and Jon discussed the issues surrounding legalities that they encountered. The team had to understand the legal restrictions and responsibilities governing data and also overcome the challenges of freedom of expression and privacy. Jon highlighted that this was a significant challenge, as the aim wasn’t to stop children from interacting digitally, but to ensure that they didn’t expose themselves to any risk. Here the team worked closely with the ethics and legal teams to ensure that the AI was tracking the right elements in conversations and that it was inclusive of all demographics
- Deliver: Once approved by all parties, the Own It app made it to the app store in September 2019.
Read more about Jon’s top six technical challenges that he and the team overcame.
Measurement, moderation and management
Jon went on to answer questions from the audience covering a whole range of topics from diversity and inclusion to success measurement and management. Key takeaways include:
- the app has a solid core audience that has been using it since the launch; the app collects user journey data which has allowed the team to refocus some areas and make it more user friendly
- In order to measure how well the app is working, the team are looking at a number of metrics on top of the amount of time young people spend on the app. These include: the number of interventions, whether the intervention turned into an action, and how many return visits the app received
- Jon confirmed that the double diamond approach to the design was good but that it also had its draw-backs due to the number of ideas generated. His main piece of advice was to keep a tight rein on what is practical and can be done within the budget and time frame
- The team wanted to ensure that the app can intervene on topics that children are worried about so topical, supportive content is added on an on-going basis. For example, supportive content surrounding the Black Lives Matter movement has recently been added
- Another challenge that the team had to overcome was bias within data; for example, the team found that if a sentence was started with “Men are”, the emotion classifier turned angry. The team realised that with protective characteristics that measure emotion, gender should not play a factor and, so, the team built a system that labelled any gender pronouns as neutral.
Want to find out more about the BBC Own It app? Download it here.